Gradientech AB is a company that develops a new ultra-rapid system for antibiotic susceptibility testing and their product will hopefully be used in health care for more efficient diagnosis of sepsis. Today the product has difficulties with the analysis of polymicrobial tests. For our project, the task is therefore to examine the market and the demand for the ability to detect polymicrobial infections. Our project is meant to provide ideas for improvements on their current method for the feasibility to detect several types of bacteria in the same test. Through literature studies, we have received an overview of how optical methods using image analysis can work and how they could be implemented in Gradientech's product. Image analysis can make the identification of bacteria possible. Interesting objects in the image can be distinguished from each other with filters that handle, for example, size, shape and light intensity. A potential improvement in the classification of bacteria by optical methods may be to test several different angles and wavelengths. This is to get more training data for machine learning and thus more accurately detect and analyze samples. The results from literature studies and interviews with clinicians, a microbiologist, and professor in image analysis have led to a conclusion that techniques that contribute to a decrease of antimicrobial resistance will be crucial for the future. In Sweden, the development of resistant bacteria is pretty low, but in many other parts of the world, there is a different reality. The product’s greatest potential will probably be in countries with a high incidence of resistant bacteria and with educated personnel available to use the product. Ultra-rapid systems for AST can contribute to saving lives by making sure that the right type of antibiotics is given directly. Detection and identification of polymicrobial samples in sepsis are also relevant, but it will not contribute to reduced antibiotic use since the patient's health will always be prioritized as of now. Finally, we reached the conclusion that identification of polymicrobial tests is not a necessary feature for the success of the product. This is above all due to the mass spectrometry method MALDI-TOF, a technique that is already well-established in health care systems. It can quickly and cheaply identify bacteria in clinical samples.

Akut- och psykiatriska avdelningar på länssjukhuset Ryhov i Jönköping använder sig av snabbtest för drogscreening med varierande kvalitet under de tider då analysinstrumentet Konelab Prime 30i inte är bemannat. Syftet med studien var att utvärdera sensitivitet och specificitet hos Multitest 15 från tillverkaren Acro Biotech, och jämföra resultat från två olika avläsningstider. Antalet urinprover som samlades in för analys uppgick till 272. Positiva och negativa urinprover med drogkoncentrationer inom ±50% från varje drogs gränsvärde insamlades. Senare inkluderades drogkoncentrationer utanför detta intervall. Proverna testades med Multitest 15 vid laboratoriet för klinisk kemi på Ryhov efter utförd analys med Konelab Prime 30i, vars analysresultat utgjorde referens. De droger som testades var amfetamin, metamfetamin, ecstasy, bensodiazepiner, buprenorfin, kokain, metadon, morfin, THC, oxykodon och tramadol. För alla droger sammantaget var sensitiviteten 86,7% - 100%, specificiteten 33,3% - 100% och träffsäkerheten 71,4% - 94,7%. Provurvalet inom intervallet ±50% från gränsvärdet var begränsat, vilket avsevärt påverkat dessa beräkningar, och Konelab Prime 30i använder semikvantitativ metod vilken endast ger approximativa koncentrationsvärden som referens. / The emergency and psychiatric wards on the county hospital Ryhov in Jönköping utilize onsite drug testing with varying quality during evenings and night-time when no staff are operating the chemistry analyzer Konelab Prime 30i. The aim of the study is to evaluate the performance of sensitivity and specificity of Acro Biotech Multitest 15 and comparing results from two different reading-times. The number of urine samples collected for analysis was 272. Positive and negative urine samples with drug concentrations within ± 50% from cut-off were collected. Later, concentrations outside of this range was included. The samples were tested with Multitest 15 at the laboratory for clinical chemistry at Ryhov after analysis with Konelab Prime 30i providing reference results. The drugs tested were amphetamine, methamphetamine, ecstasy, benzodiazepines, buprenorphine, cocaine, methadone, morphine, THC, oxycodone and tramadol. All drugs included, the sensitivity was 86.7% - 100%, the specificity 33% - 100% and the accuracy 71.4% - 94.7%. The sample selection within the range ±50% from the cut-off value was limited, which significantly affected these calculations, and Konelab Prime 30i uses a semi-quantitative method only providing approximate concentration values for reference.